InterVue

Turn interview anxiety into mastery with InterVue. Experience hyper-realistic AI voice interviews, live code execution, and real-time proctoring tailored precisely to your resume.

Comment

InterVue – AI-Powered Mock Interview Platform

Inspiration

Have you ever walked into a technical interview feeling fully prepared, only to freeze the moment you are asked to code while someone watches you?

We have.

Most interview preparation platforms today fall into three categories:

  • Too expensive
  • Static question banks
  • No real interview pressure

The reality is that studying questions is very different from experiencing an interview.

We built InterVue to bridge that gap — a hyper-realistic AI interview simulator that recreates the pressure, interaction, and evaluation of a real technical interview.

Our mission is simple:
Democratize high-quality interview preparation for everyone.

What InterVue Does

InterVue provides a complete simulated interview experience tailored to every candidate.

1. ATS Resume Analysis

Users upload their resume and target role.

The AI automatically analyzes:

  • Resume formatting
  • ATS keyword compatibility
  • Skill relevance

Users receive insights to improve their resume before the interview.

2. Adaptive Interview Generation

Instead of asking generic questions, InterVue dynamically generates interview questions based on:

  • The user's projects
  • Their skills
  • The job description

This creates a unique interview loop for every user.

3. Realistic Interview Simulation

The user enters a fully interactive interview environment.

Features include:

AI Voice Interviewer

The assistant listens, evaluates responses, and asks follow-up questions.

Live Code Sandbox

Candidates must implement solutions in a real coding environment.

AI Proctoring

Eye tracking and face monitoring simulate real interview pressure.

4. Deep Performance Analytics

After the interview, users receive a detailed performance report including:

  • Technical performance
  • Communication clarity
  • Behavioral feedback
  • Proctoring flags

The report is generated as an exportable PDF for tracking improvement.

System Architecture

Frontend

  • React 19
  • Vite
  • Tailwind CSS v4
  • Framer Motion
  • Redux Toolkit
  • Monaco Editor

This provides a fast, responsive, and modern UI experience.

Backend

  • Node.js
  • Express.js
  • MongoDB

Handles:

  • User authentication
  • Interview session management
  • Analytics storage

AI Engine

We orchestrate multiple models using the OpenRouter API.

Models used:

  • GPT-4o-mini → Resume parsing and ATS analysis
  • Qwen-2.5-Coder-7B-Instruct → Technical question generation

This multi-model architecture optimizes cost, speed, and intelligence.

Voice & Proctoring

  • Web Speech API (webkitSpeechRecognition) for voice-to-text interaction
  • MediaPipe for real-time face tracking and eye movement monitoring

Everything runs client-side to reduce server load and latency.

InterVue Scoring Algorithm

Candidate performance is calculated using a weighted evaluation model across technical and communication metrics.

$$ Final Score = \left( \sum_{i=1}^{n} (w_T \cdot T_i + w_C \cdot C_i)

\right)

(\alpha \cdot P_{flags}) $$

Where:

  • (T_i) = Technical score for question i
  • (C_i) = Communication score for question i
  • (w_T, w_C) = Role-dependent weights
  • (P_{flags}) = Number of proctoring violations
  • (\alpha) = Penalty multiplier

This scoring system balances technical ability, communication, and behavioral integrity.

Challenges We Faced

Real-Time Latency

Creating natural conversation required minimizing delay between:

  1. Speech recognition
  2. AI response generation
  3. Text-to-speech output

We optimized:

  • API call batching
  • asynchronous state management
  • frontend rendering pipelines

to prevent the AI from interrupting the candidate.

Secure Code Execution

Running a coding environment in the browser introduced challenges such as:

  • memory isolation
  • runtime crashes
  • component lifecycle cleanup

We implemented careful sandboxing and component unmount handling to ensure stable execution.

Accomplishments

We successfully integrated:

  • Conversational AI
  • A live code compiler
  • Computer-vision proctoring

all running simultaneously inside a single web application.

Achieving this without browser crashes or performance issues was a major milestone.

What We Learned

This project gave us deep hands-on experience in:

  • AI orchestration
  • prompt chaining
  • cost-efficient model routing
  • complex asynchronous browser APIs

We also learned how to balance AI capability with real-time UX constraints.

What’s Next

InterVue is only getting started.

Multi-Language Code Sandbox

Support for languages like:

  • Python
  • Java
  • C++
  • Go

Recruiter Dashboard

Companies will be able to:

  • Send InterVue links
  • Run automated pre-screen interviews
  • Review candidate analytics

VR Interview Mode

Using WebXR, candidates will be able to enter a fully immersive virtual interview room.

Share this project:

Updates

Submission history